)

ED 155 625*,

AUTHeRTITLEINSTITUTION

SPOTS /Gila

PUB DATECONTRACTNOTE

. .

IDDRS PRICE

6

DOCUMENT MBSVAI

CS. 004 142

Holland, Janes G.Analysis of behatior in Reading:Instruction..Pittsburgh Univ., Pa. Learning Research andDevelopment Center.. .

Natiobal Inst.' of EdUcation (DEER), Bashington,_D.C.Apr 76400-75-0049 .043p.; Paper presented at the COnference cn Theory andPractice of Beginning Reading Ustructior-, Uniiersity

'.of Pittsburgh, Learning Research and LeveOpmentCenter, April 1976;"Pos related documents,, see CS 004132-114, CS .004 135, CS 004 137-173, PD 125 315 andED14,5f399;.Best copy available

HP-S0.83 BC -$2.06 Plus FOstage.DESCRIPTORS Beginning. Reading; Behavioral Scietce.iesearch;

Behliioi Chaining; Conference Reports; ContingencySonageaent; *Instructional Design; Primary Education;.*Reading Instruction; Rea4irg Research; StudentBehavior; feacher.Behavior; leaching NethcdS.

ABSTRACTThe use of thebehavioral gcienee principle of

contingepcy .in several approaches to teaching beginning reading is.

discussed in this paper. The failure tc provide a contingentrelations4ip'betvein the given reading skill and student success. indesonftreting the skill.can Ivied to repeated apparent 'accesses.withdrut the student actnally manifesting the skill. Frequently,errorless progressions (shaping behavior) are used to establishheavily cued or prompted behavioral manifestations different from the

, desired terminal behavior; the cues are then gradually. faded.Altertatively, a progreksion in complexity-along the targetedbehavior can be designed. Approaches tc,beginning reading differ.principally inthe nature of the progression used. Ike examples,tbatare.givei-indicate that progression in tbs target behavici is usuallythe host effective approach. Hence, basic research cn errorlesslearning provides a frasevork for an informed judgment cn therelative serits.of different approaches too beginning reading.((Discussion. that followed the presentation of the paper ire included.)Author/EL) -

***************************************rnsevos***1**********Reproductions supplied by ZDRS are the best that can be sade *

fros the origInal.docament.*************************************wwws:***6

-7-

ti

I

JS DEie'rEsiT 06 Mt At. 714

EDtiCTos. wIELfOIEs.110*Jst. sS*,%.,,E Os

lEIStiCOs.

, -'. SEE ;FE PP:.7)(EZ.: s'". E:

-E PEPSI. Z'4, PC M'S E :6+ P

'S.D tiE PPESIN' C'. S"...EFD,:' ON P:1

Analysis o,f Ee?Isvior in F.eaM'r.z

James '.";. H-.:11and

..nlversi.v of Pittst...rgr.

Aarn:n;z

BEST COPY AVAILABLE

Ell:ming Head: Analysis of Behavior in Reaaino ILstxuction

',This paper was presented at the conference on Theory and Practice cfBeginning Reading InstructionolUniversity of Pittsburgh, Learning Researchand Development Center,;April !976..,/

Conferences supported by a grant to the L-earning Research and DevelopmentCenter from the National Institute of Education (NIE), United States._Department of Health, Education, and Welfare, as part of :::E's CompensatoryEdycation Study. The opinions expressed do not necessarily reflect theposition or policy of NIE, and no, official endorsement should be inferred.NIE Contract 0400-75-0049 2

S

4

Analysis of Behavidr iii Reading Instruction

James G. Holland

Unive;sity of Pittsburgh

My aim,in this paper is to illustrate the usefulness of the

experimental atialysAts of behavior to the design of reading in-

. atruction. From the perspective of the analysis of behavior, all

instruction, including 'instruction in 'reading, consists of ar-

ranging sequences of contingencies. each contingency has three

parts (a) some material is presented to.the student; (b) the

student interacts with the material; and (c) the interaction has

a consequence, frequently a reinforcing consequence, By concen-

trating on this deceptivdly simple concept of "contingency," I

hope to demonstrate that its implications. reach far beyond the-

"behavioral-copy-editing" level to providing a new basis for con-

trasting the modus operandi of several approaches to teaching

initial reading.

When contingencies are effectively designed, the material

presented ts appropriate .to the student's current level of&hieVe-

rsent in the subject, the interaction is a behavior that takes the

learner another step toward mastery, and the consequence rein-

forces the desired, newly emitted behavior. Instructional mate-.

rials can fail to establish effective contingencies either by

providing inappropriate cues that alloc4 the correct, answers to be

achieved by trivial student behaliiors, or by eliciting behaviors

719

3

Analysis of Behavior in Reading'Instructio

720

only superficially related to mastery. Both problems can be.

illustrated in relation to reading instruction. I will begin

with the first and probably the. most commoninappropriate cue-

ing.

1

Inappropriate Response Continzenoies.

Usually in an insteuctionalsituation only a small part of

the student's activity is public and observable by the teacher--

a question is answered about material the student has read, or

a answer is written, to a probl&m in the lesson material. The6

stud'ent's final public performance should depend upon the Cor-d

rect execution of the private act--a coerect answer should in-,

dicate that the material has been read or the problem worked

out. Over-cUeing or inappropriate cueing enables the student

to respbnd.correctly without having actually perforted the task

that the lesson was intended to evoke. For example, when a

teacher has students "read" .aloud .together, some.chikdren will

respond on the basis of cues provided by other students rather

than on the basis of the printed text.-.Such responses are "mis-

cued." The importance of insuring'appropriate response con-.

tingencies is often,unaerestimated because of the apparent 1.6-

probability ofcOntrolling,a.child's behavior by very subtle

inappropriate cues. Psychology's famous "horse" story is an

illustration of how, .given'proper training coffairtions,t)ehavior

can be controlled by inappropriate stimuli so subtle that even

the source of the sues, the horse's trainer, was not aware

giving them..4 .

Analysis of .Behavior in Rending Instr=tio

721

cMany students of general psychology are familiar with the

stbry of Clever flans, the remarkably intelligent Russian trot-

ting horse, who had (in the hands of an extremely skillful

teacher).demonstrated-an ability to count, perform arithmetic

problems, comprehend complex questions., spell, and even read

pfungst, 1911). Unfortunately, Hans could not tali so he

1

would manifest these skills by head shaking, and pointingor

by-tapping out numerical answers :with a hoof. In demonstrati:Ig-

his ability to read, a series Of words:(in Ger:ran yet) were

placed before him and he would point with'hissnose to the cor-

rect words. Hans's teacher, the remarkable Herr von Osten,

used friendly encouragement and rewards in the form o4,breads

or carrots. He worked entirely'without aversive consequences- -

and this in 1904 while Skinner was but an infant in rural east-

ern Pennsylvania:

The first group of scientists evaluating Hansr.was quite

taken in. A second group discovered that Hans could solve

problems only with an informed audience. For example, if per-

sons giving the question did not themselves knoV the answer,_

Hans was unable to solve it,. and, in a more revealing instance,

if the audience and Hans were given different questions, Hans

answered the question put to the audience. Hans had learned

to respond to subtle cues from the -audience. To determine

whether of nor Hans could read correctly, the audience had to

identify the position of the correct word and then watch Hans

'imake his choice. In doing so, the audience gave Hans the cues

ceneeded to get his carrots. YpE Herr von Osten himself

. ,zJc

.

.Analysisof Behavior in Reading Instructior,-;.

722

believed that he had trained Hans to perform on an intellect\ial

1'

__

-level equal to normal- thirteen- to fourteen-year-old ilAren.,

Indeed Herr vatvOsten hid used a careful training sequence

.On' the targeted skills he began with an extremely simplLfied)

subject matter,. For example% inIceaching Hans to "read," he,

..

first taught Halls- to indicate a single word, then one of a ,

'pair of words. He physically guided the earliest responses,

then indicated them with gross physical motions and-then Fore

subtle signs. Hewas always ready with carrots and bread at

the-instant that his student followed hiss lead. Two progres-

sions were present in this training procedure--the targeted

tasks increased in comple'xity and, at the sate tize;the ex-

traneous cue of physical guidance, jestering,' etc., decreased

gradually until the trainer himself was unaware that he was

still providing4pues. Unfortunately, the targeted cue', the

text itself; never.gained control. The fading stimuli, which

were not completely removed until the revealing experiments

were performed, continued to control.

The lesson of Clever Hans has been difficult to learn.

At the 1975 APA convention an experiment was described in

which a severely retarded girl was caught a sight vodab.dary

of 54 words or phrases (Rosenbaum & Breiling, Note 1). The

--T-reading material appeared on individual cue cards'and in-

eluded such phrases as "point to the car" or "poin to the. .

man smiling." The girl responded by pointing to o ofsev-

oral pictures on cards laid out on the table before hei. ,Dur-.

ins training,, if she,was correct she was praised and given

'6

Analysis of Behavior in P.6ading Instructic-

793

candy. .If incorrect, the,lxperimenter read the instruction

aloud himself and mode .d the behavior for the iubject. Afte:

a pause the card was presented again and the student prompted4

to do what the card said. If' necessary the subject was phy-

'1sicall.y guided through the beh'avior, Letrning appeared to be

amazingly rapid--after only eight 15-20,minute training sessions,

the severely retarded subject was close to 1007. correct.b

The experimenters were mainly interested in identifying

what reinforcing aspects of the situation.ere important.-4

Candy was fiist eliminated and various attempts made to obscure

the face of the experimenter. But none of these efforts had

more thtn a minor disruptive effect on performance until a pro-.

jector was used to present the cues 'and the experiMenter left

the room ta observe through a one-way mirror. Performance

then disintegrated to between 10% and 357. correct_ on various

sessions. It seeks to me that the ghost of Clever Hans haunts

this experiment.

When both Clever Hans, and a severely retxr6ed child tan

convince people especially attuned to the misuse of cues that

they are'reading," can we possibly be surprised by the whole-.'

sale failure.in maintaining appropriate response contingeocies

is thi traditional classroom? John Holt, aleading author in

the educational protest literature,' provides.many.examples

from classrooms of children usirig inappropriate cues--cues

which circumvent the to- be- learned skill.: In one instance

111be describes a classroom.in which the teacher was performing

a drill on the blackboard designed to *teach the identification

s' 7 4

dt

1

I

1

4 (

Andlysis of Blehaviorin Reading InsFructio:

724

. of nouns, adjectives; and verbs. She arranged 'three columns

On the blackboard, one for-each-category, and as she said each

:word she asked a,child in which cola p it belonged. , As Holt

describes it:

.

the percentage of hits was remarkably high, especially

since it was clear to me from the way the children Were

talking and acting that they hadn't 'a notion of what.

Noups, Adjectives, and Verbs were. Finally one child

said, "Miss , you shouldn't point to the answer

each tim." The teacher surprised, and asked what

. she teant. The child-said, "Well, you don't exactly

point, but you kind of stand next to the answer."

This was noclearer, since the teacher had been stand-..

ing still. But after a while, as the class went oA, I' III,

thought I saw what the girl meant. Since the teacher'

wrote each word down in its proper column, she was,

in a way, getting herself ready to write, pointing her-

self at the place where she would soon bewriting. From

the angle of her body to,the blackboard the children-qr.

picked up.a subtle clue'to the -correct answer (Holt,

1964, P. 13). )

As though this were not sufficient, this teacher kept

the three column S approximaEely.equal idlength. Thus, the

first word after the columns came into balance might be a

blind guess, but for the next there would be one chance in

two to get it right, and for the final word the correct column

was a dead give away. 8.f

Analysis of Behavior n ReadihgInstr. tior

.725

Holt describes many: such failures to assure a contingent

relation bet;4een correct,performence and actual mastery in the

classroom. It Seems likely tbat most of-us can think of a host

of examples from our own experience. The new educational tech-

nology should help the teacher who' is, frankly, faced with an

almost imposgible task. Indeed it does help, but even with

technology and with theory-inspired design. there are pzoblems

ofof inappropriate contingencies.

0. Moore's Talking T7Dewriter: One technologically

based effort in reading instriction received considerable pop-

ular attention in the 1960's, spurred by a film 'report Showing

a few children learn#g to read using a "Talking Type writer"

(Moore & Anderson: 1960), In the film, O. K. Mooreq daughter

and a few other children, under careful adult guidance, use an'

eledtric typewriter. Early in training as the child strikes

a key the adult says its letter name. Later the adult approxi-

mates the phonic sound as a key is struck. Next a projector

presents letters and the child matches the letter.by striking

the appropriate key. The adult operates a hand switch to cut,

off the power should the child begin to strike arviilappropriate

key. When the appropriate key is struck the adult pronounces

the phoneme. Later the projector presentg whole words and

still later, sentences.. The adult continues to use, the hand

&witch to put off power 'before inappropriate keys are'struck

and to pronounce the individual phoyfemes'and the completed word.

The,entertaining filmed report gives the impression of well-

"designed contingencies in a progression of tasks that result

in reading.9

aft

Ailalisis of Behavior in Reading Instruction

/26

Unfortunately, the film is not sufficienElydetailed to

.permit one to ascertain preciSely what elements of the described

procedures are effective on. whether. there are. other, perhaps

unrecognized, roles played by the adult. ;When the "talking

typewriter" was fully automated, it became possible to examine

4refuily the effects of the procedurei and,contj.ngencies..

Automated. equipment, prepared by Edison Responsive Environment,

produces letter,names 'br 4ppropriate phonemes automatically as.

.a key is struck: 'When the child is to match a projected letter,

. ,inappropriate keys are locked permitting only the'designated

key to operate.. -Richardson and McSweeney (1970) then set about

the task'of experimentally evaluating the 'talking typewriter"

. procedures, and found the results most unimpressive. anal-

yzing the failure of these procedures, Richardson and McSweeney

noted that it the initial free exploration days when.a presse'

on,any key produces the appropri?ate sound, there is, in,fact,

absolutely no contingent. relationship between the keyboard let-

ter and the spoken sound so far as 04 child is concerned. The

child does not need to'look-at the lettei nor attend, to the

produced sound, even though this phase is supposed to-reach

letter-sound-associations. Thus, many children fail to learn

these associationsi.:--

In the later phase, when the child is shown words or let-

ten, and Is to type corresponding. letters, the automated key-,

board locks all incorrect letters. The child teed merely hunt

among the keys to find the unlocked one and, in time; type a.1

perfect message without attending to ehe visual display, th0

10

.

I Analysis of 'Behaliior- in Reading Instrudtic-

727

,

ters on the keys, or the sounds produced in striking the keys.

If a. child has completed the first phase withoUt learning letter-,

sound as-sociadions, is4.especially likely that he or she yill

continue. to perform in this non - reading way. And that child

will continue to succeed! Nothing i,n this "advanced" phase

.assures that the child will attend to any grapheme-phoneme as--

sotiation. Clever Hans lives in the age of technology.'

Teaching Earl., Readin.,

Thus far I haves focused on problems in assuring appropriate

response contingencies, Yetone could design items ::hose cor-

rect solution indicated appropriate learur behavior, but which

the intended learners simply could not solve. Although such

items'assure appropriate reSponse contingencies for correct

answers, they are inadectlate as teaching items. Good materials

,insure that. the children can successfully perform the behaviors

called for by de.rising,a careful progression in- task complexity.

To devise appropriate-progressions in task complexity, wV

must undertake a behavioral analysis of the to-be-learned task

and of the.entri,behZvior of the learner. As a first step,

reading might be analyzed-as follows. CoMpetent readers rec-.

,

ognize a Urge nu:riber of words, And have vocal or subvoca be-

haviors'under control of whole words or phrases such that their.'

rate of reading is Tar too fast fOr single graphemes and pho-,

nerves to be functionally serving..as stimuli. In Contrast, non-

reading children may have substintial'speaking and listening

vocabulary but they do not yet have these reiponsesliander.control

J

Analysis of Behavior in Reading Instructio:-1

728.

of tektual stimuli% lChildren.earning to read: are confrOnted,

then, -with clusters of Visual forms that are highly similar tb

imp -another_ Yet these alMost identical forms were arranged

by thk- writer in response to th'e aquence,of sounds in which

,

the individual .units or characters are conteolled by 'a cor-,

-resoonding-souhd.. Fbr skilled readers, the resulting sequence

--controls their vocal.(or subvocal) behavior so that.th'ey will

say. the word or phrase that' the writer said. When confrontinga , --

an unfamiliar cluster of sharacterC:the reader can'fall back

on,re,spondir 'to a smaller unit than the total word by emitting

the sounds ttltrolled-by'groups of-syllables or, on ,rare oc-

casions, ever by the individual graphemes. This skill (respond-

ing to a new)cotbination,of letter _forms to produce the. sounds

similar to those produced by the authbr) is so uniformly dg-

veloped throughout the literate population that a single au-

thor may generate.a'completely new word andhaVe it quickly be-

come a part of, everyone' s reading vocabulary. Consider, for,

sample, the wort snafu.

'And'n'shafemight be the response of the hep EngliSh-

speaking child on-first.discovering that many iden.tical printed

letter forms, in the context of other letters;. thust control'

different sounds (such as long .and short vowels) and that many

-different letter forms control tile same sounds (c or k for /k/).,_

The reading teacher. then, has the task of getting-the nonreader

both to read a, large number Of,words without resorting to smaller

unit.dqcOdiig and to decipher new words despite the lack of a'

.simple phOnarte7gragieme correspondence in Ecgl.ish. The ear*

12

(

9

C

.' .

Analysis of Behavior in Reading Instruction,

729

tasks must be manageable by the child and; at the same time,

must progress toward the targeted behavior off' the proficient

reader. The hOtly debated issues among:look-say,'phonics,

and' linguistic apprOraches are' fundamentally debates on what

form of early steps are both. manage

ful in progressing toward profici

this point to describe some basitndings from the-experimental

the child and use--

ing. I, propose at

'analysis of.behavror which bear on these issues with the --p-e"

of suggesting'a 4peortical rationale useful both in Aaiding:

among approaches and in sharpening ..the execution of. particular

approaches. The -first illu-stration'of the.usefulnesdLof basic1IN

research findings in thdesign of reading instruction involves

a relatively minor flatter in sequencing..

Why bekin with/ the alphabet? .Chall (1967) ooncludes that

.teaching either letter nameAr phonic values is, a

necessary be-

; fore beginning reading ihttriaction. The basic dperant litera-

ture provides a ratio4ale for the usefulness of early alphabet

training in,one of the best known and mast fundamental findings,

the phenomenon of sti,t6lus generalization. In the typical de-

mo'nstration, a given ppint'on a stimulus dimension (for example,

a particular wavelengtkof light) is the occasion on which a

reiponse qii be intermittently reinforced and, as a cons'equence',

future presentations of the wavelerigth result in responding.

'During:a/later test phase, other points along the stimulus -di-.

mension are presented. These too will reiblt in some respond-.

ing; For example, if the test Color is fairly close in the

spectrum to the orginal trained color, it will evokeAlmost.as_

. 1 3

Analysis pf,Behhvior in heading Instructi'

730

much resPondinz, as the original stimulus. The frequency of

responding decreases with the Aistance from fhe trained stimulus, 410

but a sizable amount-of responding is evoked by stimuli that

jkae g"considerable distance away. Mdtre predise differenmtiation

/ok' of the stimulus' dimension results when ,responding ,one stimu-

`..

lus is reinforced while responses to other Stimuli even quite.

. -

close,' to it are unreinfoteed, '-or when -each 'stimulus is trained.

io a particular response (Nevin, 1973)..

Letter forms are- quite similar to one another.- In fact,

we-might say "all letter forms look alike to us" until we have

become personally acquainted, for..example, by being on a first-

name basis with them. Until some diffential response such as

the letter name is learned, the childrwould be expected to show

stimulus geheralizatiou/Tor letter forms by responding to dif-

feTent letterS as though they are the same. Although critics

of teaching letter naming are correct In contending that the

act-of naming letters is not actually a part of reading, dif-

Jetentiallyiresponding to letter shapes is a basic part of,

mature reading..

"Errorless" learning research and a comparison of three

early reading techniques. The bisic operantaiterature also

has something to-contribute by way of rationale conccrni g some

contrasting approaches to providing a suitable series of earn-

ing,tasks that introduce the novice to reading. Skinner's

earliest work emphasized the gradul shaping of a new skill and,

hence, what has been called "errorless learning.". Reading cur-

ticului approaches have always had some form of progression; to

have started with the complete refiling task-of the mature reader

li

Analysis of Beh4vior in Reading Instructio''

731

would be folly. But what'is asked of the beginner and,how do

the early tasks relate, o later proficiencies? These Ore the

issues which divide contesting approaches and to these 'issues,

I believe, recent basic research on errorless learning relates.

This,Fesearch has been attempting to determine the conditions'

necessary for attaining new discriminative control errorlesslyA .

(Sidman*& Stoddard, 1967; Terrace, 1963; Touchette, 1968).: Er-

rorless sequences.are tsually successful when the progression'

\\_' is from a prominent'stimulus to a mgre subtle stimulus-along

the Same dimension: In fact; a simple gradual progression on

the same stimulus dimension can easily carry control to the

limit of _sensory capacity. For example, Clever Hans's trainer

prominently signalled the correct response with gross gestures

earlywin training and lat'er, unknown to the trainer, maintained

very subtle physical' signalling.

A.problem arises when there is no means of establishing

initial control on the relevant' stimulus dimension. common

solution involVes pairing the target Stimulus with an irrele-

vant,.but highly prominent, stimu,lus which can initially con:-Z.

'tied the response. In CleVer Hans's case, the physical gts-t

turing cues were supposed to be faded out until Hans wes under

exclusive control of textual stimuli.. As we hAle seen, when -

the fading stimuli, were indeed completely temoved it was shown

that textual stimuli had peer gainbd control. This fading

cue technique, -common in the laboratory' and instructional de-.

sign, is frequeritly `ailed a "transfer of, stimulus control."0

Unfortunately, the technique is unpredictablesometimes it

r

15

,./..-

Analysis of Behavior inReading Instructiel

,:la2a

at.4{.?

works, other times"does not. Instructional designers often1

find'that developing a successful fading series involves a te-

dious trial and error_procedure. Occasionally, when formative

evaluation of instructional design is weak or absent, an un-

suCcessfUl fading series may survive to haunt users of the

t-tfinished - curriculum materials. Basic research on fading cuest

recentlylcompleted at the Learning Rese'arch and Development

'Center'by 1Judith Doran (1975) offers designers some 'guidelines.'

for devel9ping.successful fading series, and also offe-is. ,a ra-

f

w N\

tionale for choosing among several techniques of beginning'

readink instruction.

Doran (1975) reasoned that during a fading sequence, theisr

fadingstimulus and the target stimulus comprise a compound

stimuliis similar to the situations often studied in selective4

attention. ,Initially, only the fading stimulus controls the

response. She argues that'the criterion stimulus will gain

control Et the,end of training only if it taegins to share con-

,.

.trol with*the fading stimulus sometime during the fading se-

querits. This shared control should happen consistently only

when the-target stimulus itself is sufficiently prominent that,

the subject can begin to use it early in the fading series. If

%oEh,cues do not share control well before the fading stimulus

has beeti -dOmpleiely removed, the sequence will essentially train

finer and fpier discriminations of the irrelevant cue stimulus.

-Im Doran's experiment, children acquired a size discrimina-

tion for circ1,14 projeCted successively'on a single transillumi-

. skated kdy (seeligure 1). The positive !stimulus.(

S+

always 111

16

..4

41.

Analysis of Behavior in Reading Instructio:

733

a brightly.illuminated 14 pa.diameter circle. The child responded

Insert Figure 1 about here

6by pressing.theilluminated lc,ey. On S+trialresponding was

intermittently reinforced.

--The negative stimulus (S- ) was a,smallee circle. On S"'

trials, responding was never reinforced. During training, *.a

fading stimulus or irrelevant cue was superimposed on the S-

4

Circle. In this instance, the fading series ,began with S- as

a.comioletely dark key to which children did not respond, and.

gradually over a series of trials, the S- was brought up to

-lull brightness. Thus, the prominent irrelevaiit cue.was bright-

;Hess, and the target discrimination was circle size.

There were-4hree different problem difficulties involving

.three negative stimuli (S- ) of different sizes. In all problems

the S+ was a 14 mm diameter circle. In problem A, the simplest

problem, the S- circle diameter was 5 mm, easily discriminable

from.S . In the intermediate problem'B the S- circle diameter

was 10 mm, and-in theplifficult problem C, 12 cm. Since St

and S- ,were never present simultaneously,.a 2 mm differene

in diameter was a very difficult discrimination.

After each block of ten trials, two probe stimuli were.

used to identify which element of thecompound stimulus con-

trolled responding (Fig. 1). One.pcobe was the current S-_

circle size, but at full brightness; the Companion probe was

the S4 circle size, but at"thebrightness appropriate to that

17

Analysis of tchaytor tn Reading Instructio

734

point in the S fading series, If the subject responded to the

smaller, fully bright circle, but not to the large, dimmer circle,*

responding was controlled by the irrelevant fadir'g cue, brightness,

alone. If the subject did not respond to either probe, dual -con-

trol was indicated since, neither size nor brightness alone was

sufficient to evoke a response. Responding to the large, dimmer

circle but not to the small, bright circle indicated control by

size alone. Finally, if the subject" resppnde'd to both probes,

neither s;imulus was controlling the'response a condition which

ordinarily prevailed when the subject was responding to all S1§

in training as well.

Figure 2 shows data'for five subjects on the most difficult

Insert Figwe 2 about here.

of the size discriminations (Problem C) on.four successive

daily sessions (Cithrough CO. The data are nown as individualc

records for each of the childreri in each of the sessions:* They

. are plots of trials on which errors occurred. The horizontal

_lines in t1e first three or four blocks of trials indicate that

. performance was totally error free. The stairstep effect in the

final blocks indicates that as the brightness cue became too..

difficult and finally impossible to use, errors occurred in the

'form of responding to the S . Typically, in the early trials

consistent control by brightness iA indicated by the first,/sec-

ond, and third probes,4followed by control by neither for the'

berth and fifth probes. In the final and criterion phase,

-13

Analysis of Behavior in Reading Instructidil

735

without a brightness cue, there are extensive errors.

The cooditions represented here are analogous to, those

seen. in curriculum materials. irk which a prominent irrelevant

cue isused to assure initial correct responding, and the'ir-

relevant cue then gradually faded out. When there is no -con-

tingent'tingenr relationship for atteriing to the target dimension and

little likelihood that it will, enter ijK1 controlling rela-

tionship because of its relative lack of prominence, the fading

.sequence will frequently fail to assure the desire'd final per-.-

formancle.

In,the second condition in Doran'A study, children first

learned the easiest of the three discrimin4ions(Problem A),

then the intermediate.one (Problem B) and finally the,most.dif--.

-ficult (Problem C). This sequence involved a progression on

the relevant target eimesion, circle size,'from ea Test Prob-

lem A to the diffidult Problem C. In each individual problem,

however, the brightness fading series alustrated in rigure 1

was led. It is apparent from the data presented in Figure 3

Insert Figure.3 about here

that theiekve subjects collectively made very few errors in

learning, and only one of the subjects, Ara, showed any errorsf /

in the criterion phase of the difficult problem. Typically,

before the...criterion phase of each problem was presented,.probes

IIIindicated dither control by size or, more often, dual control.

'by siz "and-brightness.

19

sr -Analysis of Behavior in Reading Instruction

736

Because of the.striking difference in size between S+ and

S -in Problem A, brightness fading was successful. Hering-got-

ten size to control responding in\.one problem, it was easy to

establish it in tsucCessively more difficult problems. Apparently,

then, a progression\on the target dimension maintas attentionA

to the relevant stizulus'and assures success. Relying on fading

an irrelevant cue dim&nSio is a questionable practice which,

when it rust, be used, st d be paired wi;-.11\a particularly

salient targeted stimulus which can then fprther'progress

toward an increaginglY`subtle cr4erion discrimination.

Stepping Stones to Readinz. As I indicated earl.ter, ap-

proaches to establishing a useful progression in the teaching

of reading differ in ways relevant to the Doran study. The

vague feature in a reading currictium prepared at the Learning

Research and Development Center called Stepping Stones to Readinz

(Kjeldergaard, Glaser, and Frankenstein, 1973) is 4ts use of color

to code the various graphemes. A given color is associated with

a single phoneme even though the phoneme is represented by various

graphemes. For example, purple was used for the /i/ sound in

high, kite,,and f Moreover, when a given letter form is as-

sociated withdif rent. phonemes, the colOr coding functions to

distinguish them. For example, in kite, kit, and fir, the letter

"i" weacl be printed in purple, red,'or dark grin respectively.

A total of 11 colors were used, but vocabulary was choosen so. ,

that no page had more than'five colors. Hem then is a reading

program th'at has as its, maior feature the use of quite prominent

fading stimuli irrelevant to the targeted stimuli of letter-form

V .20

Analysis of Behavior in Reading Instructic:

.737

discriminations:spelling patterns, and context.. Findings from

411 tbi Doran studyciwould predict some diff?bulty with this approach.

4An experimental evaluation ofStepping Stones to Reading

was carried out by Helen Popp (1972) and setae of'her results I

find compatible with the suggestion that.this fading technique

might be ineffective. Y.er study not Only evaluatid the,stan-

dard version of'the program, but had another group use a-"re-

duced color version" which dropped the colorcoding more quickly.

T5e version which dropped the color coding =ore quickly gave

seperior posttest results. This superiority of the, faster fad-

ing held as well for a "low readiness" group:that sine investi-

gator had thought might need the mortcolor prompttng. They did

not, possibly because the color prom=pting was not helping. It

is unfortunate that no version was tried that had no,color cod-

ing. We cannot conclude from this study that color coding was

no bplp at all, but the basic data from Doran's study raises

this suspicion, and there is no evidence to dispell it.

Modified alphabet.' Another way,sto attain consistent cor.-

-respondence between graphemes and phonemes in early reading is

to modify the alphabet 'so that the printed forms are in a (more

or less) one-to-one relationship with English.phonemes

critical markings are among the most common and least extreme

Of these modificatio9s0 but one which has caught most attention

as been Sir James Pitman's Initial Te inAlphabet (ITA) con-

,

ststipg e(144 darlacters, each having a single phonemicvalue.

.

_,"""S..

r

p

Analysis of Behavior in Reading Tnstructiol

Insert Figure 4 About here

738

In the beginning Stage of reading instruction all materials are

printedin the ITA characters illustrated in Figure 4. Later,

after the children master reading this alphabet, they switch to

the standard alphabet.

To translate this into the lange of.the Doran experiment,

thekITA is a set of initially controlling stimuli ("fading stimuli")

and the standard' alphabet forms are."target stimuli." Unlike

the color-coding approach, the ITA bears some relevance to.the4 .. .

.

.

targeted standard alphabet. First, the characters of the stan-

dard alphabet also are included in the ITA, although there they

representonlyone'phoneme. Many of the new forms created for

the ITA also bear a reasonable resemblance to the standard fords

which must eventually control the child's responding, although

other ITA forms are quite different.

Since the c101d does start reading In ITA by discriminating

1

forms, ome of which are close approximations to the standard

alphabe ,the ITA-approach should be less subject to. problems

-of fading on an irrelevant dimension than a color-cading.method.

However, ITA does notfseem to be an ideal fading series even so.

Here, ii.with".;xperiments on the, color-coding approach, one

alight wish for more help from-the evaluation data., but uhfartu

nately, here too the data .are not sufficient to settle the point.

Downing.(064a, 1964b) has. shown that ITA is mote easily learned

than the standard alphabet, and hag also reported better word rec-

ognition a half-year after transfer to the standard alphabet for

I Xnalyss ofBchavior in Reading Instructic:

739

ITA children than for children trained initially on the standard

alphabet. However, Chall .(1967) itentifiessomsproblems with

these data, such as omitting to report the makeup ofdthe classes

qr to mentioncoptrolling for time spent in leading instruction.

Most significantly, shk points out that the ITA group started

phonics work "a good deal earlier" that' controls. She concludes:

So far, the experimental evidencAs still too

limiteeto allow definite'conclusions about the

long-term advantages (and disadvantages) of using

a modified alphabet. That ITA has its share of

failures'we know from a paper .[ofd Sir James Pitman. . .

We also can infer some lack ofsuccessVrem Do'..-ning's

report (1964b) revealing that after two years, 15

',,peralk of ITA-trained children had not yet been

Cran&ferred to [standard alphabet) (Chall, 1967,

p..124) .

This hint of difficulty in switching to a standard alphabet is

suggestive of a deficiency in the "fading series."

tloomfield's linguistic approach. One early reading ap-

prosth is exemplary in Using.a progression on a relevant dimen-

sion and thus it having the child perform from the beginning in

terms of the stimuli that are to..46ntrOlfinalskilled perfor-

mance..

'Bloomfield and Barnhart's (1961) Let's Read is anialpha-.

.betic approach to English. Bloomfield analyzes reading as "pro-

ducing the phonemes of one's language whenone sees written marks

Which conven tonally represent these phonemes" (Bloomfield &

23

4

IN/

Analysii of Be or in Reading Instructiol

740

Barnhart, 1961, p. 26). But early phoneme-grapheme correspon-

dence is not taught, by segmenting and subsequatly synthesizing

phonemes in what has been called the systematic phonics approach.

Instead, only whole words are used from the very beginning.' They

progression in task complexity is from short words with completely

regular spellings in which each letter has only one phonemic value'

&et, got, gun, but not gem) to more complex reading tasks,

with the complexities grouped according to their spelling char-*

Acteristics. Only in part four of ele five-part)program Oce

-many irregularities introduced. By this time the child is a

veteran at handling the code.

A potential difficulty in the Bloomfield's linguistic ap-

loroach is that it is apparently difficult to get children to

induce the sound values of letters. Nearly everyone who has

modeled a reading program c:') Bloomfield has introduced some

form of single-letter sound analysis. Given this additional ,

prpmpting, the Bloomfield system seems to be an outstandillg ex-

ample of progression widn'haintains the desired behavior thoughoui

The dirt experimental data available on Blomfield's ap-. 44,

proach is in the form of a comparison o..alinguistic program

(let's Read), a modified linguistic program (Structural Reading.,

Series) and a basal reader.(Ginn Basic Reading Series) by Sheldon,

Nichols' and Lashtnger (Note 2).-.While all three prOgramitaught

mell'and the differences in test results were minimal, the lin-

guistic ,approach did yield better- performance on ,the Gilmore Oral

Reading,Teit, and the Linguistic and modified linguistic groups

betterthan the basal.in the.Sianfard Test sUbtests on wordAll

24

Analysis 4,f, Behavior in Reading ItIstrlm,tf7

741

meaning' and spelling. Unfottunately,,eXperiments which would

directly show the effect of progression along relevant target be-

haviors,in'the context of reading programs have yet to be done:

Siifmnary

Aimple basic concept from, the experimental analysis of be-.

havior--the concept of contingency-7has been shos.../h to have im-

portant implicaCions for'classroom teaching practices, curriculum

deVelopment, and for the highly instrumented ptodUcts of eduCa-,

. ,

,tioftal technology. -Forfeach, the failure to pravide a .contin-

gent relationship between the given reading'SkilI.and student

,success in performing the xerci,se canlead to repeated apparent

success without the studentperforming the skill at all.

Frequently, errorless prOgressions first establish heavily

cueOr'prompted behaviors that are different-in kind from the-,

'desired terminal behavior. the cues or prompts are tken gradually

faded. ,Alternatively, a progression in complexity along the tar-

geted behavior can%be designed.' Approaches to initial reading

differ principally in the nature of the progression used. The

color coding of different saunds'is an extreme example of,fading

on an irrelevant dimension, the Bloomfield linguistic approach

a clear example of using almost exclusively the target behavior.

The modified alphabet approach is intermediate between'these.

Basic research on errorless learning indicates that progresiion in

the target behavior is usually the most effective approach. Hence,

.the basic research on erroless learning provides a framework

for an informed judgment on the relative merits of the several ap-

proaches to beginning reading, and guidance for future develop-

ment of initial reading. material.

25

Analysis of Behavior in Reading Instructie

742

That the afheoretical, descriptiVe approach of.the science

of behavior shoUld see easy application to reading instrudtion 4111

is hardly surprising. Skinner (1954) long ago palled attention

to the implications of an experimental analysis of behavior for

educational practice,. With this approach basic work leads na-

turaity to application in practice, and experience in practice

leads equally naturally to basic analytiC research. Indeed, I

'have shown elsewhere (Holland, .1976)'that basic work in errorless

discrimination learning received' its firsts-impetus from common

practice in the early days of programmed instruction.

The ease of application of the experimental analysis of

behavior contrasts markedly with the generally meager_ practical'

yield of*theoryloriented research. This contrast is a direct

result of the difference objectives and methods of the two

types of research. Theory as the term is used here, is an ex-

plaiation of observed events by events on a different levele,of

inalysis. Commonly, errors or latency of responses are the data

explained by theories involvin physiological or mentaltevents.

These theoretical,entitids typically are the principal object of

study for the theorist. ThUs cognitive theorists might be study=

14 "Memory stores" and the."accessing.of memory". To do tflis

they may measure latencies in reading words as an indirect indice

of the non-behavioral entity which is the/ object of study. A

-systematic understanding of reading or even of latencies in read-

ing words,is not the direct object.of the research. Appliqation,.

if it is to ever occur, must await the;eventual completion and

verification of the theory.

26--,

Analysis of Behavior in Reading Instructi:

743

In contrast, the analysis of behavior determines controllingA

rentionships on the behaviora). level.% The "pure"'laboratory

study may use arbitrary-stimuli and the complexity of "theworld

of practice may be reduced for analysis, but the laboratory and

practical setting are not different in kind. Errorless learning'

is the same phenomenon whether seen in reading or in arbitrarily

chosen laboratory material. IA, the experimAtali#alysis of be-

havior "applied vs. pure" research distinction disappears..

A thorough-going experimental;analysib,of reading instruction

could improve our understanding Of reading and solve many of the

problems' of reading instructfbn.A

-.0

27

4

Analysis of Behavior in Reading Instructic:1

744

Figure Captions

figure 1. Size and Luminance of S+, S and Probe-

Stimuli.. 1

Figure 2. Performance of five subjects given four size

discrimination problems in the Sequence CCCC. Each problem

had a 50-triil pogramwed series, a 10-trial' criterion: and

- ' 5 probe sets. For the programmed series and the criterion,

each trial moves the data line one step to the right, each

.errorful trial aliso moves the line one step upward. Perfor-

mance on probe sets is coded as t, b, s, d. The probe key

identifies the probe esponding and the indicated stimulus

control for eactt code lettefr.

Figure 3. Performance oflive subjects given four size

discriftitation problems in the Sequence ABCC. Each problem

had,a 50-trial programmed .series,

5 probe sets. For the programmed

each trial moves. the data line one

errorfliittial alio moves .the line

a 10-trial criterion, ''and

series and the criterion,

step to the right, each.1

one step upward. Performance

bn probe sets is'coded as n, b, s, d. The probe key identifies.

the probe responding and the~ indicated stimulus control for

each code letter.

Figure 4. Sir .Tficias Pitman's Initial Teaching Alphabet.

Note. From the book Initial Teaching Alphabet by Sr: James Pitman.

Copyright...1973 tty Pitman Publishing Corp. Reprinted by permis-

Bien of Pitman-Publishing

I28

S

Ahalysis.of Behavior, in Reading Instruction

745

Reference Notes

41. Rosenbaum,- M.'S:, & Breiling, J. The development and

functional control of reading comprehension behavior.

Paper presented to the American Psychological Association

Convention: 1975.

2. Sheldon, W. D., Nichols, N..J., & Lashinger, D. R. Com-

parison of three methodsoffragiing reading in the seeorld

grytel 1USOE Cooperative Research Project Report No,'3231.

Syracuse: -University of Syracuse, 1967.

IT

Y

AndIYsis of Behaviof in Readir Instructtc1

746

References

Bloomfield, L.-, ..Ec Barnhart, C. 1.. Let's read: A linguistic

approach. Detroit: Wayne State University Press, 1961.

Chali, J. 'Learning_ to read: The great debate. New York:

McGraw-Hill, 1967.

-Dorhn, N. J. Control by stimulus featuees during.,fadinz. Un-

published doctor dissertation, University of Pittsburgh,

1975.

ADownitg, J. A. The i.t.a (initial teaching alphabet) reading ex-

periment. The Reading teacher. 1964a, 18; 105-109.

Downing, J. A.-;.,The i:t.a. reading' experiment: Three lectures

on the research in infant schools with Sir James Pitman's

.initial teaching alphabet. London: Evans-Brothers, Ltd.,

1964b. is

43

Holland, J. C. Reflections on the beginnings of behavior analysis0

in instruction. L. E. Fraley and.E. A. Vargas.' Proceed-

'ings of the:Third-National Conference on Behavior Research .

44 304

Analysis of Behavior'n Reading Instructio

747

so

and Technology in Higher-Education, Atlantic Ca.: Georgia

State University, 1976.

Holt, J. How children fail. New York: Delta Publishing Com-,

r

pany, 1964.

Kjeldergaard, P. M., Frankenstein, R., & Glaser, R. StepoingStones

to Reading: EmDerimentai Edition. New Cerftury/Appleton-

.

Century- Crofts, Inc., 1969.-

_ --

Moore, 0. K., & Anderson, A. R. -Early reading and writing. Pitts-

4.

buigh: Basic Education,- 1960. Sixolken-millimeter color and

sound notion picture.

Nevin; J. A.- Stimulus control. In J. A. Nevin & G. S, Reynolds

(Edt.), The study of behavior:. Learning, motivation alkd in-

stinct. Glenview, Ill.: Scott; Foresman, 1973.

Pfungse, 0. Clever Hans. New York: Henry Holt and'Company,

1911.

Popp. H. M. Test project -for the LRDC beginning reading program

%

"Stepping Sedhes to Reading." fittshurgh: University of

31

Analysisof Bohavior:in Reading Instructioal

74t

Pittsburgh, Learning Research and,Development center, 1972.

(Publication 1972/15)

Richardson, E., & McSweeney, J. An analysis_of the E.R.E. Talk-

k.

ing TYpewrsiter" as a device for teaching beginning reading

skills. Educational Technology, 1970, 10(2), 81-88.

Sidman, M., & StOddard, L-. T. The effectiveness of fading in

programming.a simultaneous form discrimination for, retarded

. children. Journal of the Experimental Analysis of Behavior;

1967; 10, 4715,111.

Skinner, B.F. The science of learning and the art of teaching.

Harvard_ Educational Review, 1944, 24,' 86-97.

'Terrace/H. S. Discrimination learning with and without "errors."

Journal orthe Experimental Analysis of Behavior,. 1.963, 6,

1-27. .

Touchette, P. E. The effects of graduated stimulus change on the

da.

acquisition of a simple 4scrimination in'severely retarded

boys. Journal of`the Experimental Analysis of,Behavior, 1968,

11, 38-48.32

el

0

April :/ti--P .M. 749

,OPEN DISCUSSION OF HOLLAND PRESENTATION

NUT: I would like to call your attention to some- research going on at

.

Rockefeller University by Ray McDermott, who is doing studies Of video tapes of/

reading classes. He has Determined that one of the major distinctions .between

the good readliiiii group and the poor reading group is the fact that the children. -

in the good reading group are, unconsciousbr, telling the teacher who to call on

next. In other words, they seem to be controlling the behavior of the teacher

through their body movements.

STICHT: Jim, regarding, the ITA, I frequently found that the criticism was that

even though initially it might. have been a more rapidly effective decoding

program, three or four years later it made no difference. Do those tests that

come. three or four years later foius more on comprehension than on decoding, and.."

is there any reason to think that if it took one year to teach decoding by one

program, and: a-half a year by then4ber, and itocomprehension tests given three

s later indiCate that decoding has been learned under either one,ne, would -you

sell sect the faster decoding program?

HOLLAND: I as talking about thq immediate effect.

-S,STICHT: But I mean should we choceeLa-docoding program because it is fast? How

A

can decoding effect comprehension three years later, unless comprehension itself

is somehow taught, too.

IMAM: 3peaker requested that his comments be deleted.

33i

'aa

i

'2is

April 13 --P.M. L, 750

STICH?: Sometiles I have seen wrAte-ups about the ITA that say' it's rejected

. because although it enables people to learn decoding faster, three yeart down th%

road there is no difference In the children.

But why would you expect there to be a difference in children, three years

latir, if they have, by then, all learned to decode? Why would anybody look for

a diffdrence then in a comprehension task?

HOLLAND: Speaker requested that his comments be deleted.

STICHT: I don't either.

HOLLAND: Speaker requested that his comments be deleted.

SAMUELS: In regard to the warning you shared with the group about using dolor

coding as an' aid to reading, I agree that the problem is one of transfer of

stimulus control from the color to form and shape. The color isactually used as

a prompt to ekcit the response, and the student is supposed to focus attention

,on the visual characteristics of the word or letter. An experiment we did at

Minnesota same years ago points out the problems. There was a very simple task.

gave college students a paired associate list tolearn. The paired associate

',-

list had. a high degree of visual atimOlus siailarity, and the subjects had to

book up words to the printed stimuli. One stimulus wee printed inyed; and all

Of the _other stimuli were printed in black, but the shapes of tbe stimuli were

We told the stuilents to ignore the color and to focus attention on

Asps. During the training phase, we always got the correct response to the.

stimulus in red, but, on transfer, weoxiiminated the color, and even though the.

subjects got repeated warnings and repee)ted trials, whenever'tbe transfer phase

34

-1

April 13--P.M. ,751

occurred anithe color was eliminated, we got no response. It is obvious that

the college students were unable to ignore the patent cc/Mr °me d toattend to-.

shape.

HOLLAND: They were worse off, really.

SAMUELS: Oh, yes, on the transfer task, they were worse off. It points up tie

'problem 'with _these programs that have color aid. The student-keeps on looking

at the irrelevant color dimension, and neverfocusses on the form, which is the

relevant cue for the task of reading.

HOLLAND: I introduced Doran's study by talking about the selective attention

literature. Tou are, correctly, carrying the analysis a step further than my

paper did. Trittni4 on a single component of a compound stimulus will sake

acquisition sore dyficult for the other component. Extensive pr4training with

an irrelevant cue retatiii learning the criterial task. Some fading sequences do

just that. It's not simply that they do no gbod, they Mai actually do harm.

BATEMAN: Jay, on the same point, is it Lsportant to distinguish, in discussion

of the use of color, the way it's commonly done frog the way it's done -iw those

few programs that merely use color as a cue to say the word?

SAMUELS: I think you have the same problem. If the student has a potent cue to

use for the Correct 1esponse, the prindiple of least Wort operates; the

student will use that cue which most easily elicits the correct response; in

this eas: it's color. Hut in real reading, there are no cotors. So at the

transfer phase in real readint when the color cue is roved, you are worse off,

35

April 13 --P.M.

and I think that's theepoint that To was making:

752

Do you want to ha4e rapid initial learning, but poor transfer, or would you

rather trade thab off foe a little slower initial training, but good transfer?

STICK: That's not exactly what I had in mind, but it's close: What I had in"i

mind is this: Sometimes we hear a lot about differant ways of teaching decoding,

and different approaches. One of thoeilwas the ITA approach, for whidh you gave

al nice theoretical rationle-for why it ought to work. One criterion you might

evaluate a program by is how rapidly it brings about the same effect that another

approach will bring, about: Now, the point I want to get to is that we have to

haiii some-criteria for seleting approaches. Rate, that is, how quickly it

achieves it night- be one of the deciding "atprs. gut if we are going to wait

until three or four yealls later, and then say, °Well, children who used ITA

learned to decode faster than did-the other children, but three or four years

later the effects washed out' I want to know what washed out? Maybe using, one/

kind of model, you could say that if ybu had given the person the ability to

`unlock the key to that spoken languages that if they couldn't speak too well or

comprehend 'spoken language too well, haven't gotten the key real fast, wouldn't

sake any difference three years later, if they still didn't build. and drill

heavily on that. You end up with a normal distribution again-thrie years later.

s

BOLLAJ Speak ar requested that hVb emits be deletqd.

;

SIMI: I get the isprvIsion when you are talking about this other paper that

.

you tend to favor this natural contingency. Could you give us an idda of what

that paper 'pup be like?

I

O

April 13 --P.M. 753

HOLLAND: Speaker requested that his commenta bi deleted.

./

GOODMAN: I am interested in Tour relating either your task or your- methodology

to some theoryof language development, for instance, Skinner's verbal learning

theory.

HOLLAND: That's a large order, for a brief 'discussion.

. .

GOODMAN: Can I rephrase it then? Is there anything special about language or

*verbal learning that makef the kinds of things you have been talking about more

or less appropriate; or is that just like all'other kinds or learning.

HOLLAND: Language behavior has the name basis as other behaviors including

reittnilL. Verbal behavior is set apart only in that it involves a reinforcing

calamity.

POSNER: This evaluation of ITA is a good place for analyses. Coding is real'l'y

quite important, because if you look fdr the visual to phonetic decoding, ITA

'looks like 1a good thing to do, but if you place stress on the orthography,

particularly as Massara and Venezky did, and particularly if you think that

getting.s good orthographic chunk depends, in part, on visual familiarity, then

maybe is really a quite disastrous thing tb do, even for the purpose. of

decoding. 4

HOLLAND: Good point.

. CHALL: I think st the present time the data show that ITA has more problems for

37

April 13 754

.spelling than for reading,

BATEMAN: Would you clarify or specify the objective

.Bloofield-Barnhart is a viod way to get to?

HOLLAND: Getting kids to have the right kind of word attack.

BATEMAN: Speaker requested that her comments be deleted.

which you think

HOLLAND: Okay. When the child is through Bloomfield and Barnhart, the child can

read Just about any word that he can understand.

BATEMAN: In your paper do you have any citations to data for that?

HOLLAND: I base this on having used the material to teach two children to read.

The effectiveness data I cite in the paper are from Sheldon, Ichols, and

Lashinger. (But note the qualifidation I made regarding the difficulty in

getting children to induce sound values.)

BATSMAN: Would you modify the methodology? I realize the materials are well

programmed, Out did yOu do what the little front pait says to do?

HOLLAND: More or less. The:front part teems to talk "as if you have to move much

slower than I think is desirable, It seemed to me going fast is the important

thing.

RESNICK: In my memory of Bloomfield- Barnhart, all of the concerns that Beck and

38

41

April 13--P.M. 755

'Block raised last night about the Palo Alto sequence would hold: it would 0.

interfere with the skillslof scanning ahead to figure out what the graphemic

context was, and also it would probably provide relatively little natural

reinfor4ement for a ldng time.

HOLLAND: 'Speaker requested that his comments'be deleted.

RESNICK: And the talking typewriter?

HOLLAND: Speaker requested that his comments be deleted.

IMUELS: Thin morning Barbara Bateman was talking

the teacher point out the...elevant dimensions of,

AI/ A, 2 and g different? They are visually similar,,

relation to each otber.,gibson points out that

,relational and have to be seen that way.

6 r.

about the importance of having

let's say, letters. How are 12,

out you have to see them in

distinctive features. are always.

Now, earlier you said that Skinner would recommend that when you teach a

child to recognize letters, you might want to hive ap errorless discrimination

pattern. Well, that would almost point to the opposite set of letters.

For example, if you never wanted a child to confuse . with another letter,

you would put:b and Z next to one another. It would be very easy to discriminate

one tram the other.

Sow would you devise a program which ;Would teach a child' what the

distinctive features of letters such as,., A, 2 and-11, lower case, or lower case

ja, ji,./ and mg They hair* similar, features; they are related- to -one another.

.Sow would yoU devise a program that would teach distinctive features, so that the

. 39

April 13--P.M. 756

child could rapidly identify.one letter and not mix it up with another?4

HOLLAND: It happens that I have collaborated with Sid Bijou many years.ago on

teaching machine program which established just such a skill. We weren't working

on letters of the alphabet at all. Ra6er we were attempting to establish a

(we might ca.1.1 them "new aptitudes" at LRDC today). We werebasic aptitude

`teaching the space faCtor--one of Thurstone's primary mental abilities. We used

gradual progression that started-very close to your a4. The children matdhed

forms, at first with very dissimilar forms, and later with a sample and five

matching stimuli with four of them reflections of mirror images of the sample,

andinother,one, the correct alternative, was simply a rotation of the sample.

.,.The problem was for the child to find the rotated sample and reject the mirror

images. It's a difficult task, but)with a proper progression they were able to

do it. Then, as a tour de force, we used a transfer test, in- which we used

letter forms such as 11., 12.. 2, and so forth. The transfer was very good.

It did involve an errorless progression that went from easy to.difficult.

SAMUELS: Would you want to train on nonsense figures or directly on the letters

of the alphabet?

BOLLARD: If we had been primarily interested in'the alphabet, wd,would have been

trained on the alphibet; but we weren't:.

GLASER: I am not sire the message, Jim, that you want us to take home, about- v

whether or not fading, Dr tragafer of !stimulus control through fading, is

reocomended is the design of instruction in reading or not.

April 13P.M. 4

HOLLAND: Avoid fading on an irrelevant dimension even at high cost. If it

*67

-wowbe avoided, use the irrelevant )dimension early, get rid of it as fast as

possible, and make the target dimension is prominent as possible early in

training.'.

Aliot

PERFETTI: Speaker requested that -his comments be deleted.aos

HOLLAND: I wouldn't tilink so. That's very much a part of reading, you wouldn't

-want to avoid it.

PERFETTI: To take it a step further, is there any point in the levelqpment of

reading, that you want to worry aboutf-

-0 4

HOLLAND: Speaker requested that.hie comments be deleted.

PERFETTI: No, no', I will try again. There is one analysis of reading which is

under the control of something other than fors; it iS analogous to your analysis

of poorly engineered programs. That is to say, that in some sense those stimulus

factorsif we want to call them that--are not going to transfer necessarily to

the kinds of stiuluscontols that most people have in mind when they say

someone can decode. Now, I only wanted to get you to say whether you thought

there was some point at which it made sense not to worry about Chit; whether in

general you could tell, in looking at a Qhild'arrlding behaviot when-he was

,using miscues in your sense, as compared to the sense in which he is really doing

samethinigseffthandipanktgrul in the reading situation.

4

HOLLAND: If I understand, you art asking whether a child, who because of

41

4

)April 13 --P.M. 758

content, says "up" instead of,down,* is using miscues.

PERFETTI: Let's take a tefinite arrticle example that comes out.. That's one

extreme. Suppose he says,. "Jimmy caught the bus," instead of "caught a bus."

That is something most of us would. think probably is not worth correcting, sinc<

he. is on the right track.. There are other kinds of miscues which would be

farther'off. my problem is seeing a stimulus control analysis mapping onto that

type of behavior observatio thinking about what use to make of it. .--

11,

HOLLAND:, If you viewed reading as being totally and exclusively decoding, I

Bless you might worry about that, but I don't know anybody that believes that.

4

PERFETTI: No, the examples get more interesting. The "example is, *He caught the

. hue,* instead of .*caught the train.* Some people would say that is not So far

off.

BOLL*: Again, if your foCus is entirely or decoding, I think you would correct ,

: But that is not a decoding emphasis, Jim. That is an emphasis on

There is a difference in meaning'betweem *He caught the bus" and He

caught the train.* They Just happen to hibe comparable messages in some context.

RiSNICK: Is your question: Should we always beseeking errorless learning, or

close to errorless learning, or are there times whmm the errorsirre functional?'

Or am I oversimplifying? .42-

April 13--P.M. 759

PERFET1I: No, it it much simpler. It is just a question of trying to decide

Uhen, if you are going to use stimulus control language, you talk about behaviors

connected with the print on the page, some reading behaviors are gang to look

pretty fir-off base, they are attached to the wrong stimulus conditions. The

fading problems that you-nave pointed to have,. perhaps, 'been built into the

instructional program too severely. I am simply asking whether you have any

further insights into this problem that exist in so many other *forms. How"do you

know when to worry about it apd when not to worry about ft?

HOLLAND: The problem Is in the specification of the target behavior. I don't

think everybody's definition of reading would include worrying about contextual

misreadings of occasional sorts, but I suspect that such overconcern would result

in a poor quality of reading--slower and more ponderous reading.

PERFETTI: Presumably that's part of the answer. I mean, the history of the

overall reading behavior is a clue.

J

END SESSION

43